U.S. patent number 7,035,836 [Application Number 10/115,887] was granted by the patent office on 2006-04-25 for method and apparatus for controlling a vehicle suspension system based on sky hook approach.
This patent grant is currently assigned to STMicroelectronics S.r.l.. Invention is credited to Riccardo Caponetto, Olga Diamante, Giovanna Fargione, Antonino Risitano, Domenico Tringali.
United States Patent |
7,035,836 |
Caponetto , et al. |
April 25, 2006 |
Method and apparatus for controlling a vehicle suspension system
based on sky hook approach
Abstract
A method for controlling a vehicle semi-active suspension system
comprising at least one suspension, providing for: detecting
vehicle dynamic quantities during the vehicle ride; using the
detected dynamic quantities, determining an index of ride comfort
and an index of roadholding; applying a weight factor to the index
of ride comfort and to the index of roadholding and, based on a Sky
Hook control model, determining a target damping force
characteristics for the at least one suspension of the suspension
system; controlling the at least one suspension to put the
respective damping force characteristics in accordance with the
calculated target damping force characteristics. The weight factor
is calculated dynamically during the vehicle ride, by means of a
fuzzy calculation on the detected vehicle dynamic quantities.
Inventors: |
Caponetto; Riccardo (Catania,
IT), Diamante; Olga (Siracusa, IT),
Risitano; Antonino (Aci Catena, IT), Fargione;
Giovanna (Catania, IT), Tringali; Domenico
(Augusta, IT) |
Assignee: |
STMicroelectronics S.r.l.
(Agrate Brianza, IT)
|
Family
ID: |
8184474 |
Appl.
No.: |
10/115,887 |
Filed: |
April 3, 2002 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20020185827 A1 |
Dec 12, 2002 |
|
Foreign Application Priority Data
|
|
|
|
|
Apr 4, 2001 [EP] |
|
|
01830233 |
|
Current U.S.
Class: |
706/47; 706/14;
706/900 |
Current CPC
Class: |
B60G
17/018 (20130101); B60G 2400/102 (20130101); B60G
2400/252 (20130101); B60G 2500/10 (20130101); B60G
2600/1879 (20130101); B60G 2800/70 (20130101); Y10S
706/90 (20130101) |
Current International
Class: |
G06N
5/02 (20060101) |
Field of
Search: |
;706/47,900,14
;701/37 |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
39 30 517 |
|
Jan 1991 |
|
DE |
|
0 499 790 |
|
Aug 1992 |
|
EP |
|
0 538 965 |
|
Apr 1993 |
|
EP |
|
1 018 445 |
|
Jul 2000 |
|
EP |
|
1 063 108 |
|
Dec 2000 |
|
EP |
|
Other References
Angela K. Carter, Transient Motion Control of Passive and
Semiactive Damping for Vehicle Suspensions, Jul., 15, 1998,
Virginia Polytechnic Institute and State University, Thesis, 1-104.
cited by examiner .
Riccardo Caponetto, A Soft Computing Approach to Fuzzy Sky-Hook
Control of Semiactive Suspension, 2003, IEEE,
786-798.quadrature..quadrature.. cited by examiner .
Jukka-Pekka Hyvarinen, The Improvement of Fully Vehicle Semi-active
Suspension through Kinematical Model, 2004, University of Oulu,
Finland, Dissertation, 1-157 .quadrature..quadrature.. cited by
examiner.
|
Primary Examiner: Hirl; Joseph P.
Attorney, Agent or Firm: Jorgenson; Lisa K. Bennett, II;
Harold H. Seed IP Law Group PLLC
Claims
What is claimed is:
1. A method for controlling a vehicle's semi-active suspension
system including a suspension coupled to a non-suspended mass, the
method comprising: detecting vehicle dynamic quantities during a
ride of the vehicle; using the detected dynamic quantities,
determining an index of ride comfort and an index of roadholding;
applying a weight factor to the index of ride comfort and to the
index of roadholding and, based on a Sky Hook control model,
determining target damping force characteristics for the suspension
of the suspension system that are suitable to substantially
minimize vehicle body acceleration and to substantially minimize a
variation of a force transmitted by the suspension to a road
surface; and controlling the suspension to put respective damping
force characteristics in accordance with the determined target
damping force characteristics; wherein the weight factor is
calculated dynamically during the vehicle ride using the detected
dynamic quantities; and wherein determining the target damping
force characteristics includes minimizing a weighted sum of the
vehicle body acceleration and of a variation of a force exerted by
the non-suspended mass onto the road surface, the weighted sum
being based on the dynamically calculated weight factor.
2. The method according to claim 1 wherein the weight factor is
calculated using a fuzzy calculation performed on the detected
vehicle dynamic quantities.
3. The method according to claim 2 wherein dynamically calculating
the weight factor includes: fuzzifying the detected vehicle dynamic
quantities; applying a prescribed set of fuzzy rules to the
fuzzified quantities; and defuzzifying a result of said rules to
obtain the weight factor.
4. The method according to claim 3, further comprising using the
calculated weight factor to determine damping coefficients of the
Sky Hook control model of the suspension.
5. The method according to claim 4 wherein determining damping
coefficients of the Sky Hook control model of the suspension
comprises determining a first damping coefficient of a first damper
coupled between a wheel and a body of the vehicle, and determining
a second damping coefficient of a second damper coupling the body
of the vehicle to an inertial reference system, the first and
second damping coefficients being related to each other by the
calculated weight factor.
6. The method according to claim 5 wherein determining the first
and second damping coefficients comprises keeping the second
damping coefficient fixed, and determining the first damping
coefficient on a basis of the calculated weight factor.
7. The method according to claim 5, further comprising calculating,
on a basis of the first and second damping coefficients, a value of
force to be exerted by the suspension onto the vehicle.
8. The method according to claim 7, further comprising determining,
on a basis of the calculated value of force exerted by the
suspension on the vehicle, drive signals for correspondingly
controlling the suspension.
9. The method according to claim 1, further comprising dynamically
calculating respective weight factors for each suspension of the
suspension system.
10. An apparatus for controlling a vehicle's semi-active suspension
system including a suspension coupled to a non-suspended mass, the
apparatus including a control system adapted to receive detected
vehicle dynamic quantities during the vehicle ride, the apparatus
comprising: means for determining an index of ride comfort and an
index of roadholding on a basis of the detected dynamic quantities;
calculating means for applying a weight factor to the index of ride
comfort and to the index of roadholding and, based on a Sky Hook
control model, for determining target damping force characteristics
for the suspension of the suspension system that are suitable to
substantially minimize vehicle body acceleration and to
substantially minimize a variation of a force transmitted by the
suspension to a road surface; and control means for controlling the
suspension to put respective damping force characteristics in
accordance with the determined target damping force
characteristics; wherein the calculating means dynamically
calculates the weight factor during the vehicle ride using the
detected dynamic quantities; and wherein the calculating means for
determining the target damping force characteristics includes means
for minimizing a weighted sum of the vehicle body acceleration and
of a variation of a force exerted by the non-suspended mass onto
the road surface, the weighted sum being based on the dynamically
calculated weight factor.
11. The apparatus according to claim 10 wherein the calculating
means comprises a fuzzy controller to dynamically calculate, on a
basis of a prescribed set of fuzzy rules applied to the detected
vehicle dynamic quantities, the weight factor.
12. The apparatus according to claim 11 wherein the calculating
means receives the detected vehicle dynamic quantities, calculates
the weight factor and determines, on a basis of the calculated
weight factor, damping coefficients of the Sky Hook control model
of the suspension.
13. The apparatus according to claim 11 wherein the calculating
means comprises a computation unit to calculate, on a basis of the
damping coefficients, a target value of force to be exerted by the
suspension onto the vehicle, and a look-up table to determine, on a
basis of the calculated target value of force, drive signals to
correspondingly control the at least one suspension.
14. A semi-active vehicle suspension system coupled to a
non-suspended mass, the system comprising: means for detecting
dynamic quantities during a vehicle ride; means for determining an
index of ride comfort and an index of roadholding on a basis of the
detected dynamic quantities; calculating means for applying a
weight factor to the index of ride comfort and to the index of
roadholding and, based on a Sky Hook control model, for determining
target damping force characteristics for a suspension of the
suspension system that are suitable to substantially minimize
vehicle body acceleration and to substantially minimize a variation
of a force transmitted by the suspension to a road surface; and
control means for controlling the suspension to put damping force
characteristics of the suspension in accordance with the determined
target damping force characteristics; wherein the calculating means
dynamically calculates the weight factor during the vehicle ride
using the detected dynamic quantities; and wherein the calculating
means for determining the target damping force characteristics
includes means for minimizing a weighted sum of the vehicle body
acceleration and of a variation of a force exerted by the
non-suspended mass onto the road surface, the weighted sum being
based on the dynamically calculated weight factor.
15. The suspension system of claim 14 wherein the calculating means
comprises a fuzzy controller to dynamically calculate, on a basis
of a prescribed set of fuzzy rules applied to the detected vehicle
dynamic quantities, the weight factor.
16. The suspension system of claim 15, further comprising
exploiting genetic algorithms to optimize the fuzzy controller.
17. A method for a vehicle having a plurality of wheels, each of
the plurality of wheels being coupled to the vehicle by a
suspension having variable damping characteristics, the method
comprising: measuring dynamic quantities associated with operation
of the vehicle, including speed, vertical speed with respect to a
surface upon which the vehicle is traveling, vertical speed with
respect to each of the plurality of wheels, roll, pitch and turning
angle; calculating a value for a weight factor for each of the
plurality of wheels, using fuzzy computation; calculating an index
of ride comfort and an index of road holding for each of the
plurality of wheels; calculating a target damping characteristic
for each of the plurality of suspensions using a Sky Hook control
model, based, in part, upon the weight factor and the ride comfort
and road holding indices, the target damping characteristics being
suitable to substantially minimize vehicle body acceleration and to
substantially minimize a variation of a force transmitted by the
suspension to a road surface; and varying the damping
characteristics of each of the plurality of suspensions according
to respective calculated target damping characteristic, wherein
calculating the target damping characteristics includes minimizing
a weighted sum of the vehicle body acceleration and of a variation
of a force exerted by the wheels onto the road surface, the
weighted sum being based on the dynamically calculated weight
factor.
18. A method for controlling a vehicle's semi-active suspension
system having at least one suspension including a non-suspended
mass and a spring between the non-suspended mass and a road
surface, the method comprising: detecting vehicle dynamic
quantities during a ride of the vehicle; using the detected dynamic
quantities, calculating target damping force characteristics for
the at least one suspension of the suspension system suitable to
substantially minimize a vehicle body acceleration and to
substantially minimize a variation of a force transmitted by the at
least one suspension to the road surface, said calculating being
based on a Sky Hook control model; and controlling the at least one
suspension so as to put their respective damping force
characteristics in accordance with the calculated target damping
force characteristics, wherein said calculating target damping
force characteristics include minimizing a weighted sum of the
vehicle body acceleration and of the variation of the force exerted
by the non-suspended mass onto the road surface, said weighted sum
being based on weight factors, said weight factors being calculated
dynamically during the vehicle ride using the detected dynamic
quantities.
19. An apparatus for controlling a vehicle's semi-active suspension
system including a suspension, the suspension having a
non-suspended mass and a spring between the non-suspended mass and
a road surface, the apparatus comprising: a control system adapted
to receive detected vehicle dynamic quantities during a ride of the
vehicle, said control system having: means for calculating target
damping force characteristics for the at least one suspension of
the suspension system based on a Sky Hook control model applied to
the detected dynamic quantities, the target damping force
characteristics being suitable to substantially minimize a vehicle
body acceleration and to substantially minimize a variation of a
force transmitted by the at least one suspension to the road
surface; and control means for controlling the at least one
suspension to put respective damping force characteristics of the
at least one suspension in accordance with the calculated target
damping force characteristics, wherein said means for calculating
the target damping force characteristics include means for
minimizing a weighted sum of the vehicle body acceleration and of a
variation of a force exerted by the non-suspended mass onto the
road surface, said weighted sum being based on weight factors, and
said means for calculating the target damping force characteristics
being structured to calculate the weight factors dynamically during
the vehicle ride using the detected dynamic quantities.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method for controlling a vehicle
suspension system, and to an apparatus suitable for actuating the
method. In particular, the invention concerns a method and a
related apparatus for controlling the damping force characteristic
of shock absorbers in a semi-active vehicle suspension system,
based on the Sky Hook control theory.
2. Description of the Related Art
The suspension system of a vehicle is intended to perform several
functions, such as sustaining the vehicle over the road or, more
generally, the ground, keeping the vibrations transmitted to the
vehicle body (for example, in the case of a car, the passenger
compartment or car body) as low as possible, distributing the
forces arising from accelerations, due for example to an increase
or a decrease in the vehicle speed and/or to changes in the vehicle
ride direction.
Several types of suspension systems have been proposed, which can
be grouped in three main categories: passive suspension systems,
active suspension systems and semi-active suspension systems.
In passive suspension systems, the shock absorbers have a fixed
damping coefficient. In active suspension systems, the shock
absorbers have a variable damping coefficient, which can be varied
continuously under the control of a control system, for example by
means of suitably controlled hydraulic pumps. Active suspension
systems can thus perform the above-mentioned functions adapting to
the particular ride conditions.
In semi-active suspension systems, similarly to active suspension
systems, the damping coefficient of the shock absorbers can be
varied continuously under the control of a control system, so as to
adapt to the particular ride conditions. However, while in active
suspension systems it can be necessary to supply external energy to
the shock absorbers to control the damping force characteristic
thereof, this is not so in semi-active suspension systems, wherein
the control is only directed to properly dissipating the energy of
the shock absorbers.
Semi-active suspension systems represent an intermediate solution
between passive and active suspension systems, providing better
performance than the former without being so expensive as the
latter.
The behavior of a passive suspension system including one
suspension can be determined using for example the De Carbon model.
Such a model, depicted in FIG. 1, is a system with two degrees of
freedom, and is for example suitable to represent one fourth (that
is, one wheel) of a four-wheels vehicle such as a car. The model
system includes a suspended mass 1, of mass M, representing the
mass of the car body, and a non-suspended mass 2, of mass m,
representing the mass of the wheel. Neglecting the damping effect
of the tire, the non-suspended mass m is coupled to ground (the
road surface r) by a spring 3 of rigidity kp, corresponding to the
tire rigidity. The suspended mass 1 is coupled to the non-suspended
mass 2 by means of the suspension, which comprises a spring 4 of
rigidity k and a shock absorber 5 having a constant damping
coefficient Crel.
Applying the D'Alembert principle to the model system of FIG. 1,
the following mathematical model of the suspension can be derived:
M{umlaut over (z)}.sub.b=-k(z.sub.b-z.sub.w)-C.sub.rel( .sub.b-
.sub.w) m{umlaut over
(z)}.sub.w=k(z.sub.b-z.sub.w)-k.sub.p(z.sub.w-h)+C.sub.rel( .sub.b-
.sub.w) where z.sub.b is vertical coordinate of the suspended mass
1 (the car body) with respect to an arbitrary reference level,
z.sub.w is the vertical coordinate of the non-suspended mass 1 (the
wheel) and h is the height of the road surface r with respect to
said reference level. The second time derivative of z.sub.b, i.e.,
the vertical acceleration of the car body, can be adopted as an
index of ride comfort assured by the suspension: the lower the
vertical acceleration of the car body, the higher the ride comfort.
The force exerted by the non-suspended mass 2 (the wheel) onto the
road surface r can be adopted as an index of roadholding: the
higher the force exerted by the wheel onto the road surface, the
higher the car holding of the road. Alternatively, the variation in
time of the force exerted by the wheel onto the road during the
vehicle ride can be adopted as an index of roadholding.
The limitations of the passive suspension system stems from the
fact that only one parameter, i.e., the damping coefficient Crel of
the shock absorber, is available for adjusting the two indexes of
comfort and roadholding. Since the two requirements are independent
from each other, and since the minima of the two indexes are
achieved for different values of the shock absorber damping
coefficient Crel, the system does not have an optimum solution, and
merely a trade-off solution can be found.
In principle, this problem can be solved by increasing the number
of system parameters, that is, making the shock absorber damping
force to depend on more than a single parameter. One way to do so
is represented by the Sky Hook approach.
In a suspension system based on the Sky Hook approach the force
exerted by the shock absorber onto the car body is proportional to
the absolute speed of the car body with respect to an inertial
reference system, and to the relative speed between the car body
and the wheel.
Still in principle, as the inertial reference system either the
earth or the sky can be taken. However, since the suspended mass
cannot be connected to the earth, the sky is chosen as the inertial
reference system and the suspended mass is ideally assumed to be
hooked to the sky. The corresponding system model is depicted in
FIG. 2, where s indicates the sky inertial reference system and 6
denotes a shock absorber of damping coefficient Csky connecting the
suspended mass 1 to the sky s.
A Sky Hook damper is merely an ideal device, since it is clearly
not possible to couple the suspended mass 1 to the sky. In the
practice, a Sky Hook suspension can be implemented by replacing the
shock absorber 5, having a fixed damping coefficient Crel, with a
shock absorber 50 having a variable damping coefficient, and
providing a feedback control from the car body 1 to the shock
absorber 50, thus obtaining the model depicted in FIG. 3.
Applying again the D'Alembert principle to the system depicted in
FIG. 3, the resulting mathematical is the following: M{umlaut over
(z)}.sub.b=-k(z.sub.b-z.sub.w)-F.sub.am m{umlaut over
(z)}.sub.w=k(z.sub.b-z.sub.w)-k.sub.p(z.sub.w-h)+F.sub.am where Fam
is the force exerted by the shock absorber 50 on the car body 1.
The force Fam which, as previously mentioned, must be proportional
to the absolute speed of the car body 1 with respect to an inertial
reference system and to the relative speed between the car body 1
and the wheel 2 is given by:
F.sub.am(t)=C.sub.rel(t)V.sub.rel(t)+C.sub.sky(t)V.sub.abs(t)=C.sub.rel(
.sub.b- .sub.w)+C.sub.sky .sub.b having indicated as Vrel the
relative vertical speed between the car body 1 and the wheel 2, and
as Vabs the absolute vertical speed of the car body 1. The time
dependence of the damping coefficients Crel and Csky has also been
explicitly shown.
It follows that two parameters are now available for controlling
the suspension, that are the damping coefficients Crel and
Csky.
The Sky-Hook control technique can be implemented both in active
and in semi-active suspension systems. Since, as mentioned before,
in a semi-active suspension system, differently from active
suspensions systems, no external energy is supplied to the
suspension system but rather the energy of the suspension system
itself is dissipated in a controlled way, in a semi-active
suspension system the shock absorber 50 applies no force to the car
body 1 when such a force should be opposite to the relative speed
of the car body 1 with respect to the wheel 2.
Consequently, while in both the active and semi-active suspension
systems is: F.sub.am=C.sub.rel( .sub.b- .sub.w)+C.sub.sky .sub.b
for F.sub.am( .sub.b- .sub.w)>0 the semi-active suspension
system has the following additional limitation: F.sub.am=0 for
F.sub.am( .sub.b- .sub.w)<0
Conventional Sky Hook suspension control methods provide for
choosing the pair of parameters Crel and Csky in such a way as to
find a trade-off between the contrasting requirements of minimizing
the car body vertical acceleration, so as to maximize the comfort
index, and minimizing the variation of the force exerted by the
wheel on the road surface, so as to maximize the index of
roadholding.
A weight factor p is determined which is used to weight the two
contributes; by introducing the weight factor p, the function to be
minimized becomes: F.sub.opt=p(M{umlaut over
(z)}.sub.b)+(1-p)F.sub.gnd where by Fgnd the variation of the force
exerted by the wheel onto the road is indicated.
Once a value for the weight factor p has been chosen, the values
for the damping coefficients Crel and Csky can be univocally
determined by minimizing (i.e., searching the minimum) the function
F.sub.opt. The choice of the value for the weight factor p
determines the type of driving style; changing the value of the
weight factor p, either the ride comfort or the roadholding can be
privileged.
Up to now, in the implementation of the sky Hook control approach
in semi-active suspension systems the value of the weight factor p
has been fixed a priori, and the values for the damping
coefficients Crel and Csky univocally determined on the basis of
the value of the weight factor p by using conventional control
systems, like P-I-D (Proportional-Integral-Derivative)
controllers.
BRIEF SUMMARY OF THE INVENTION
In view of the state of the art described, an embodiment of the
present invention provides a new method, and a new related
apparatus, for controlling a vehicle suspension system, capable of
providing better results compared to the conventional control
method and apparatus.
According to an embodiment of the invention, a control method
comprises: detecting vehicle dynamic quantities during the vehicle
ride; using the detected dynamic quantities, determining an index
of ride comfort and an index of roadholding; applying a weight
factor to the index of ride comfort and to the index of roadholding
and, applying a Sky Hook control model, determining a target
damping force characteristics for a suspension of the suspension
system; controlling the suspension to put the respective damping
force characteristics in accordance with the calculated target
damping force characteristics.
The weight factor is calculated dynamically during the vehicle
ride, using the detected dynamic quantities.
In a preferred embodiment, the weight factor is calculated by means
of a fuzzy calculation on the detected vehicle dynamic
quantities.
Also according to the invention, a control apparatus is
provided.
The control apparatus comprises a control system adapted to receive
detected vehicle dynamic quantities during the vehicle ride. The
control system comprises: means for determining an index, of ride
comfort and an index of roadholding on the basis of the detected
dynamic quantities; calculating means for applying a weight factor
to the index of ride comfort and to the index of roadholding and,
based on a sky Rook control model, determining a target damping
force characteristics for the suspension of the suspension system;
control means for controlling the suspension to put the respective
damping force characteristics in accordance with the calculated
target damping force characteristics.
The calculating means calculates the weight factor dynamically
during the vehicle ride, using the detected dynamic quantities.
In a preferred embodiment, the calculating means comprises a fuzzy
controller that dynamically calculates the weight factor, on the
basis of a prescribed set of fuzzy rules applied to the detected
vehicle dynamic quantities.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS
The features and advantages of the present invention will be made
apparent by the following detailed description of an embodiment
thereof, illustrated merely by way of a non-limiting example in the
annexed drawings, wherein:
FIG. 1 depicts a model of a passive suspension system;
FIG. 2 depicts an ideal model of a suspension system based on the
Sky Hook approach;
FIG. 3 depicts a practical model of the Sky Hook suspension system
of FIG. 2;
FIG. 4 is a schematic view of a suspension control apparatus
according to the present invention;
FIG. 5 shows possible membership functions for the input and output
variables of a fuzzy controller of the control apparatus of FIG. 4,
in an example of a suspension system including a single
suspension.
DETAILED DESCRIPTION OF THE INVENTION
Referring to FIG. 4, reference numeral 31 is used to identify a
vehicle, for example, a car, having a car body 32 and a suspension
system 33 comprising, for example, four wheels with respective
suspensions (not shown in detail).
The car 31 is equipped with sensors (not shown) capable of
detecting dynamic quantities such as, for example, the car speed,
the steering angle, the car roll, the pitch, the vertical
displacement, the shock absorber elongations. Sensor signals S1,
S2, . . . , Sn generated by such sensors are supplied to a
suspension control system comprising a fuzzy controller 34, a
computation block 35 and a look-up table 36.
The fuzzy controller 34 receives the sensor signals S1, S2, . . . ,
Sn and, on the basis of the detected quantities, dynamically
determines, by means of a fuzzy computation, four values p[1] . . .
p[4] for the weight factor for each of the vehicle suspensions.
Such values are supplied to the arithmetic computation block 35.
The arithmetic computation block 35 is also supplied with the
sensor signals S1 and S2, carrying information relating to the car
body absolute vertical speed with respect to the road (ground), and
the car body relative vertical speed with respect to the wheels. On
the basis of the weight factors p[1] . . . p[4] determined by the
fuzzy controller 34, of the car body absolute vertical speed with
respect to the ground and of the car body relative vertical speed
with respect to the wheels, the arithmetic computation block 35
calculates, for each of the suspensions, the value of the force
Fam[1:4] (i.e., the force exerted by a suspension on the car body)
using the previously reported formula:
F.sub.opt[1:4]=p[1:4](M{umlaut over
(z)}.sub.b)+(1-p[1:4])F.sub.gnd[1:4] where F.sub.opt[1:4]
identifies four functions to be minimized (one for each suspension)
and Fgnd[1:4] identifies the variation of the force exerted by the
each one of the four wheels on the ground.
In particular, the arithmetic computation block 35 calculates, for
each of the four functions Fopt[1:4], the respective minimum,
thereby determining four pairs of values (Crel, Csky), one pair for
each suspension. In an embodiment of the invention,
root-mean-square (RMS) values of the car body vertical acceleration
and of the variation of the force exerted by the wheels onto the
ground are used; preferably, the RMS values are normalized.
By way of example only, the following table reports possible pairs
of values of the damping coefficients Crel[i] and Csky[i]
corresponding to different values of the weight factor p[i], in
respect of one of the four suspensions:
TABLE-US-00001 p[i] Crel[i] (Ns/m) Csky[i] (Ns/m) 0 2,360 1,540 0.1
1,640 2,770 0.3 923 5850 0.5 615 6000
The calculated four pairs of values (Crel, Csky) are used by the
arithmetic computation block 35 to determine four values Fam[1:4],
representing the target force that each suspension should exert on
the car body, using the previously reported formula:
F.sub.am[1:4]=C.sub.rel( .sub.b- .sub.w[1:4])+C.sub.sky .sub.b
The values Fam[1:4] and the car body relative vertical speed with
respect to the wheel are supplied to the look-up table 36 which, on
the basis of these values, determines the electrical control
quantities suitable to drive the electrically controlled
suspensions, which can be for example fluidodynamic suspensions or
magnetorologic suspensions.
A practical implementation of the fuzzy controller 34 will be now
described by way of example; for the sake of simplicity, the single
suspension system of FIG. 3 will be considered.
The input variables to the fuzzy controller 34 can be the
following: in1: vertical acceleration of the suspended mass 1 (in
m/s.sup.2); in2: relative vertical speed of the suspended mass 1
with respect to the non-suspended mass 2 (in m/s).
As shown in FIG. 4, the output variable of the fuzzy controller 34
is the weight factor p, that is the ratio between the damping
coefficients Crel and Csky. Assuming by way of example that the
damping coefficient Csky is kept constant at a prescribed value,
the force Fam is given by:
F.sub.am(t)=C.sub.rel(t)+C.sub.skyV.sub.abs(t) and the output
variable out of the fuzzy controller 34 is the damping coefficient
Crel (in Ns/m).
The Applicant has observed that by keeping the damping coefficient
Csky constant at a prescribed value, the control logic can be
simplified. In the choice of the value for Csky the Applicant has
observed that it is better to choose a relatively high value: the
Applicant has in fact observed that the contribution of the damping
coefficient Csky to the overall damping force depends on the
absolute vertical speed of the suspended mass, which becomes high
only at the resonance frequencies, and at such frequencies the
shock absorber must develop a higher damping force. Suitable values
for the damping coefficient Csky which the Applicant has
experimentally obtained in the case of a four-wheel vehicle, such
as a car, are approximately 5800 5900 Ns/m for the front wheels,
and 4550 4650 Ns/m for the rear wheels. However, neither these
values nor the choice of keeping the damping coefficient Csky
constant are to be intended as limitative for the present
invention; greater control flexibility is achieved if also the
value of Csky is determined dynamically, instead of being kept
constant.
A suitable set of membership functions implemented by the fuzzy
controller 34 for the fuzzification of the two input variables in1,
in2 is depicted in FIG. 5. Membership functions mf1, mf2 and mf3
determine the fuzzy values of the input variable in1, assumed to
range from 0 to 3 m/s.sup.2; membership functions mf4, mf5 and mf6
determine the fuzzy values of the input variable in2, assumed to
range from 0 to 1 m/s. By way of example, all the fuzzy functions
are gaussian.
The fuzzy controller 34 can for example use the following set of
rules: R1: IF(in1 IS mf1) AND (in2 IS mf4) THEN (out IS C1) R2: IF
(in1 IS mf1) AND (in2 IS mf5) THEN (out IS C2) R3: IF (in1 IS mf1)
AND (in2 IS mf3) THEN (out IS C3) R4: IF (in1 IS mfg) AND (in2 IS
mf4) THEN (out IS C4) R5: IF (in1 IS mfg) AND (in2 IS mf5) THEN
(out IS C5) R6: IF (in1 IS mfg) AND (in2 IS mf6) THEN (out IS C6)
R7: IF (in1 IS mf3) AND (in2 IS mf4) THEN (out IS C7) R8: IF (in1
IS mf3) AND (in2 IS mf5) THEN (out IS CS) R9: IF (in1 IS mf3) AND
(in2 IS mf6) THEN (out IS C9) where C1 to C9 are membership
functions for the output variable out of the fuzzy controller
34.
A possible set of membership functions C1 to C9 for the output
variable out is shown in the rightmost diagram of FIG. 5. In the
shown example, the membership functions C1 to C9 are crisp
values.
In order to determine the crisp value for the output variable out,
the fuzzy controller 34 performs a defuzzification process, for
example, adopting the centroid or the barycenter methods.
It is to be noted that the gaussian shape of the membership
functions mf1 to mf6 is not a limitation, and other shapes could be
used, for example, trapezoidal.
Also, the output variable out, instead of crisp values, could be
defined by membership functions of different shapes, such as
gaussian or trapezoidal.
Based on the value of the variable out calculated by the fuzzy
controller 34, the arithmetic computation block 35 determines the
value of the force Fam; this value is supplied to the look-up table
36, which generates the electric control signals suitable to drive
the suspension. As already mentioned, different types of
suspensions can be used, for example the magnetoroligic suspensions
or the fluidodynamic ones, if necessary changing the drivers
thereof which act under control of the look-up table 36.
Advantageously, the fuzzy controller 34 can be optimized by using
genetic algorithms which, as known, represent an optimization
method based on the Darwin's natural evolution principle. According
to this method, within a population in continuous evolution, the
individual who best adapts to the environmental constraints
corresponds to the optimal solution of the problem to be
solved.
In the implementation of the genetic algorithm method for
optimizing a fuzzy controller for a car suspension system, the
overall acceleration of the car body has been taken as an index of
performance (target function to be optimized). In order to compare
the RMS (Root Mean Square) value of the car body acceleration to
the RMS value of the variation of the force exerted onto the road
surface, for each one of the four wheels (the so-called corners)
the force variations have been divided by the suspended mass
related to said corner, for example 300 Kg.
The target function to be optimized is the following:
Ob=aN(acc)+bN(rdh) where a and b are two constant parameters. For
example, assuming that a slightly sport behavior is desired for the
car, the parameters a and b can be respectively equal to 0.3 and
0.7.
N(acc) is a dimensionless quantity representing the normalized
acceleration of the car body: N(acc)=n1N(vert)+n2N(pitch)+n3N(roll)
where N(vert) is the normalized vertical acceleration, N(pitch) is
the normalized pitch acceleration and N(roll) is the normalized
roll acceleration. n1, n2 and n3 are three coefficients the choice
of which depends on considerations relating to the higher or lower
effect of either one of the three accelerations on the comfort.
Suitable values are for example n1=0.4, n2=0.5 and n3=0.1.
N(rdh) is a dimensionless quantity representing the normalized
roadholding, given by: N(rdh)=.SIGMA.qiRMS(Ti)/300 where RMS(Ti),
i=1 . . . 4, are the normalized RMS values of variation of the
force exerted onto the road surface by the four wheels, and qi are
four weight factors for weighting the four RMS value contributions,
one for each wheel. For example, it can be taken qi=0.25 for each
of the four wheels.
The total number of variables is 84: for each one of the four
corners 21 variables exist, which are the two inputs in1, in2 with
three gaussian membership functions (which are characterized by two
quantities: mean and width), the nine fuzzy rules, the nine
singletons for the output out. For simplicity, only the nine
singletons can be optimized, obtaining the values depicted in FIG.
5.
The fuzzy controller 34, the computation block 35 and the look-up
table 36 can be practically implemented using a microprocessor.
Although the present invention has been disclosed and described by
way of an embodiment, it is apparent to those skilled in the art
that several modifications to the described embodiment, as well as
other embodiments of the present invention are possible without
departing from the spirit or essential features thereof, as defined
in the appended claims.
* * * * *